Rank of J is deficient

Hi everyone
when running identification command I got the following output

==== Identification analysis ====

Testing prior mean
Evaluating simulated moment uncertainty … please wait
Doing 231 replicas of length 300 periods.
Simulated moment uncertainty … done!

All parameters are identified in the model (rank of H).

The rank of J (moments) is deficient!

[Eta_Gl,Eta_Gh] are PAIRWISE collinear (with tol = 1.e-10) !
[Eta_tk,Eta_I] are PAIRWISE collinear (with tol = 1.e-10) !
[rho_Gl,rho_Gh] are PAIRWISE collinear (with tol = 1.e-10) !
[alpha_Gh,alpha_Gl] are PAIRWISE collinear (with tol = 1.e-10) !

rho_I is collinear w.r.t. all other params!
rho_tk is collinear w.r.t. all other params!

==== Identification analysis completed ====

97.1% of the prior support gives unique saddle-path solution.
0.0% of the prior support gives explosive dynamics.
2.9% of the prior support gives indeterminacy.
Smirnov statistics in driving acceptable behaviour
rho_Gh d-stat = 0.350 p-value = 0.000
rho_Gl d-stat = 0.370 p-value = 0.000
alpha d-stat = 0.602 p-value = 0.000
psi_h d-stat = 0.283 p-value = 0.000

Smirnov statistics in driving indeterminacy
rho_Gh d-stat = 0.340 p-value = 0.000
rho_Gl d-stat = 0.360 p-value = 0.000
alpha d-stat = 0.584 p-value = 0.000
psi_h d-stat = 0.275 p-value = 0.000

Starting bivariate analysis:

Correlation analysis for prior_stable
No correlation term with pvalue <0.001 found for prior_stable

Correlation analysis for prior_unacceptable
[rho_Gh,rho_Gl]: corrcoef = -0.459
[sigma_c,ksi_w]: corrcoef = -0.421

Correlation analysis for prior_indeterm
[rho_Gh,rho_Gl]: corrcoef = -0.459
[sigma_c,ksi_w]: corrcoef = -0.421

What does it mean?
What would you suggest?
Thank you

Without the mod-file it is hard to tell. The general explanations can be found at ifk-cfs.de/fileadmin/downloads/events/conferences/monfispol2011/RATTO_IdentifFinal.pdf

Here is the mode file

Please have a look at it

Thank you again

Sincerely yours
data_model.xls (47 KB)
rica.mod (12.6 KB)

It means that those parameters cannot be identified, because the observables do not provide sufficient information.
For example, the two processes

Gh = rho_Gh*Gh(-1) + (1-rho_Gh)*Y + Eta_Gh; // high government spending dynamics Gl = rho_Gl*Gl(-1) + (1-rho_Gl)*Y + Eta_Gl; // low government spending dynamics
are completely indistinguishable when not observing Gh or Gl. You cannot estimate them separately.

Thank you Sir

I will remove these two equation and replace them by only this G=rho_g*G(-1)+Eta_G. maybe this will resolve the issue

What do you think Sir Jpfeifer?

best regards

That will help for these two equations. There may be another issues that may require additional steps. After replacing the equations, run identification again.

1 Like

Dear jPfeifer

Thank you for your help
As you suggest there is really an issue in these two equations then I removed them!
Now Can you have a look on the following mode file. Rank of J is still deficient! I don’t know Why
May you interpret the graph resulting from the identification and the sensitivity analysis! What really going on with this model?

thank in advance!
SA.mod (12.7 KB)